View Full Version : XMS3200 need help bad

10-18-02, 03:31 PM
I recently upgraded my CPU(look in the sig), the new processor is super stable and I have it overclocked, running 166FSB and 10.5 on the multiplier, temps under load never get past 39C, my problem is when I bench mark using PCMark 2002, here are my scores:

Thunderbird 1.333
CPU Score 3707
Memory Score 3730
HDD Score 1209

XP 1600+ overclocked to 1.750(166X10.5)
CPU Score 5260
Memory Score 3554
HDD Score 1286

The memory is running 200 pints slower, what am I doing wrong?
The voltages are at default

10-18-02, 03:39 PM
sure the timings are the same? Also sometimes scores vary in points...usualy +-25 not +- 100 though....check your timings

10-18-02, 03:40 PM
timings? in the bios, I will write what I have and let you know

10-18-02, 03:42 PM
Probably using pcmark. Does sandra also show a lower score?

10-18-02, 03:51 PM
Here are the BIOS setting I have(who knows if I did them right)

Advanced Bios Features
CPU internal Cache-Enabled
External Cache-Enabled
CPU L2 Cache ECC Checking-Disabled

Advanced Chipset Features
DRAM Clock/Drive Control
System Performance-TURBO
Current FSB Frequency-166MHz
Current DRAM Frequency-166MHz
DRAM Clock-by SPD
DRAM Timming-by SPD
SDRAM Cycle Length-2
Bank Interleave-4 bank
Precharge to Active(Trp)-2T
Acitve to Precharge(Trast)-5T
Active to CMD(Trcd)-2T
DRAM Burst Length-4
DRAM Queue Depth-4 level
DRAM Commnd Rate-Auto
DCLKL Timing-Auto
DCLKO timing-Auto
Fast R-W Turn Around-Enabled
Continous DRAMM request-Enables
Wirte Recovery Time-2T

Frequency/Voltage Control
Auto Detect DIMM/PCI Clk-Enabled
Spread Spectrum-Disabled
CPU Clock-166
CPU Ratio-Auto

VCORE 1.750
DIMM 2.50

10-18-02, 04:50 PM
If you are using the free version of pcmark 2002 you can't change the test options that means it runs 25 seperate tests for the memory score including both L1 and L2 caches, graphics memory, AGP bus and the system memory.

10-18-02, 04:59 PM
I have a registered version, I was just wondering why the memory performs so much worse with the XP 1600+ then it did with the Thunder 1.333, I even have the 1600+ running 166FSBX10.5

10-18-02, 06:15 PM
Software that is testing many different things may produce different results with different cpu's(xp-athlon,multipler,fsb),bios settings,video card drivers,agp bus settings,video card clock speeds,operating system settings,hard disk fragmentation,free memory,tsr programs,flaws in the testing software.
To give you a simple answer as to why is not easy.